planning

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill's Research & Analysis phase (references/research-phase.md) explicitly instructs the agent to read and analyze public GitHub repositories, pull requests, issues/discussions (via "GitHub Analysis" and repomix --remote), which means it ingests untrusted, user-generated third-party content that can influence planning decisions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly calls for runtime fetching of GitHub repositories (e.g., repomix --remote https://github.com/mrgoonie/human-mcp) to generate codebase summaries that are injected into the agent's context and can directly control prompts/behavior.